Fiscal Federalism, Good Governance, And Economic Growth in Mexico Maite Careaga and Barry R. Weingast* July 2001

1. Introduction

Good governance and appropriately designed institutions are now recognized as necessary for

economic growth. Yet we know too little about this relationship. Why do some governments

protect property rights, provide a stable macroeconomy, and have limited taxes and corruption,

while others are corrupt kleptocracies that prey on their citizens?

In recent years, a new political economy literature has emerged that studies these

questions by investigating the interaction of political institutions and economic performance.1

Proponents of this approach provide a general if abstract answer to the above questions: A

country’s political and institutional framework determines whether government officials face

incentives for good or bad governance. As Stiglitz (1998,5) suggests, “misaligned incentives can

induce government officials to take actions that are not, in any sense, in the public interest.”

*Ph.D. Candidate, Department of Political Science, Stanford University; and Senior Fellow, Hoover

Institution, and Ward C. Krebs Family Professor of Political Science, Stanford University. The authors thank Lee Alston, Alberto Diaz-Cayeros, Rui de Figueiredo, James Fearon, Barbara Geddes, Stephen Haber, Robert Inman, Roger Noll, Yingyi Qian, Antonio Rangel, James Robinson, Gerard Roland, Jean-Laurent Rosenthal, Ernesto Stein, and William Summerhill for helpful comments and Juliana Bambaci for research assistance.

1. See, for example, Acemoglu and Robinson 2000, Alesina and Drazen 1991; Alesina and Spolare 1997, Besley and Coate 1998; Bolton and Roland 1997; Cox and McCubbins 2001; Cremer and Palfrey forthcoming; Dixit and Londregan 1995, Drazen 2000, Inman and Rubinfeld 1997, Jin, Qian, and Weingast 1999; Knack and Keefer

2

Unfortunately, too little is known about how institutions systematically affect the incentives of

government officials and what types of institutions lead to good governance and better economic

performance.

We add to this new literature by exploring questions of governance and growth in the

context of the global trend toward greater decentralization, particularly within developing

countries. A long tradition in economics emphasizes the advantages of federalism for good

governance. More recently, proponents of market-preserving federalism suggest that an

appropriately structured decentralization fosters economic development.2 Yet recent theory and

evidence cast doubt on the assertion that decentralization promotes political and economic

development. Many scholars suggest that federalism leads to more inflation or corruption; others

emphasize that spillovers, common pool problems, and problems from soft budget constraints

result in efficiency losses associated with decentralization.3

Following Shah (1997) our approach helps explain this puzzle about federalism by

suggesting that there is not one kind of decentralization. Federal systems differ enormously in

the ways they allocate money, power, and authority across levels of government. Some federal

arrangements are therefore likely to foster corruption and inefficiency while others foster

economic growth.

We explore how variations in two institutional features – the fiscal system and electoral

competitiveness – affect subnational government (SNG) performance. In particular, we study

how variations in the distribution of locally raised taxes and increases in electoral

1995; Levy and Spiller 1994; North 1990; Persson and Tabelini 2000; Poterba and von Hagen 1999; Rodrik 1999, Roland 2001, Stiglitz 1998; Sturzenegger and Tommasi 1998.

2 For the long-standing economists’ argument, see Hayek (1939) and Tiebout (1956); for modern reviews, see Rubinfeld (1987) and McKinnon and Nechyba (1997). On market-preserving federalism, see McKinnon (1997), Montinola, Qian, and Weingast (1995), Weingast (1995) and Zhuravskaya (2000).

3

competitiveness affect SNG choices. In our model, an SNG allocates its resources across two

categories of policies: public goods that foster markets; and the public provision of private

benefits, such as corruption and subsidies to interest groups, that hinder markets. Subject to its

budget constraint, the SNG chooses between the two activities to maximize its utility, which is

positively related to each type of policy. We define good governance as the provision by SNGs

of market-enhancing public goods instead of non-productive transfers and corruption. Good

governance is therefore endogenous.

We analyze two comparative statics. The first concerns α, the proportion of locally

generated taxes captured by the SNG: ceteris paribus, an increase in α increases good

governance. The reason is that choosing public goods has two benefits: directly, it provides

citizens with utility; indirectly it increases tax revenue and so relaxes the SNG’s budget

constraint. In contrast, corruption and non-productive transfers benefit only those citizens

receiving the rents.

The second comparative static concerns the degree of electoral competition. Ceteris

paribus, electoral competition increases good governance. Although officials in all SNGs need

political support to survive, those in an SNG that faces competition must appeal to the median

voter whereas those in noncompetitive SNGs do not need as strong appeals to voters.

Our first comparative statics result has an important implication for revenue sharing

systems that we call the fiscal law of 1/n: in a federation of n SNGs, where the central

government collects and distributes all tax revenue (i.e. α= 0), the indirect benefit for an SNG of

fostering markets is on the order of 1/n that of a fiscal system in which SNGs are 100% financed

through locally generated revenue (i.e. α =1) (see Inman 1988 and Weingast, Shepsle and

3 See, e.g., Dillinger and Webb (1999), Garman, Haggard and Willis (2000), Hzou (1998), Inman and

4

Johnsen 1981). The reason is that revenue sharing creates a common pool problem. Under

complete revenue sharing, an SNG bears all the costs of providing a public good, but captures

only 1/nth of the revenue benefits. The rest of the benefits are spread across other jurisdictions

through the central government’s division of the common pool. Therefore complete revenue

sharing systems greatly diminish a SNG’s incentives to provide market-fostering public goods.

We call revenue sharing agreements in which SNGs give up their policy and fiscal autonomy

fiscal pacts with the devil (FPWD) because they increase corruption, provide fewer public goods

that enhance growth, and diminish citizen welfare.

We apply our approach to federalism in Mexico. Although 20th century Mexico has

always had a federal structure, the distribution of authority and fiscal resources among the levels

of government (and therefore α) has changed over time. We identify four phases. During the first

period, from the Revolution (1910-1917) to 1940, the central government was unable to enforce

parts of the Constitution, allowing a variety of standard common pool problems. In the second

period, 1940-80, the center became stronger and policed common pool problems. Indeed, early

in this period, Mexico was characterized by market-preserving federalism, a type of federalism

associated with economic growth (Weingast 1995). Yet Mexico continued to centralize, coaxing

the states to join a revenue sharing system. By 1980, Mexico was highly centralized, largely

federal in name only. After the government achieved centralization, Mexico was characterized

by the fiscal law of 1/n, producing efficiency losses. Of course the debt crisis and the collapse of

oil prices also affected growth during this period. Finally, since 1994, Mexico has undergone

some decentralization and liberalization. After 1994, both α and the degree of electoral

competition increased. For this reason, Mexico offers a natural experiment of the isolated effects

Rubinfeld (1997), Jones, Sanguinetti, and Tommasi (1999), McKinnon and Nechyba (1997), Prud’homme (1995), Rodden (2000), Sanguinetti (1994), Stein (1998), Treisman (1999a,b), and Wildasin (1997).

5

of α over good governance (until the late 1980s), and then of the combined effects of α and

political competition (since 1994).

Consistent with our theory about federalism and good governance, Mexico’s pattern of

growth corresponds to these phases.4 In the first, real GDP growth was slow, averaging 1.4

percent per year. In the second, often called the “Mexican Miracle,” Mexico sustained four

decades of rapid growth, 6.8 percent per year. In the third, growth slowed considerably to just

2.3 percent per year. In the last period, 1994-present, growth has improved somewhat: real GDP

fell over 6 percent during the peso crisis (late 1994 and 1995), but then has grown at an average

of 5.1 percent per year (1996-99).

We provide several types of evidence supporting our predictions for the case of Mexico.

First, with respect to the differential willingness of states to join a fiscal pact that limits their

ability to tax, our approach predicts that the more market-oriented states should be less likely to

participate. We show that this holds in practice. Second, the theory predicts several changes in

the behavior of SNG governance after they join the FPWD. Decreases in α should lead SNGs to

spend more on rents and less on public goods. We provide some evidence of the changes in the

patterns of expenditure after states joined the pact. The fiscal law of 1/n combines with the

common pool problem created by revenue sharing to predict that state taxes should fall and total

federal taxes should increase following the 1980 change in the fiscal system. We show that this

happened.

Third, we calculate α implied by the various revenue sharing formulas used since 1980.

The initial formula, made explicit in 1980 but employed previously, returned to states exactly

what they contributed. Because the federal government also spent considerable sums in the states

4The quality of data from the early periods is poor, so these figures must be treated as approximate.

6

at this time, we calculate that α was on the order of .2. We also show that the current formula,

dating from 1995, has somewhat better properties, allowing states to receive on the order of one-

quarter of any increase in revenue generated in their state. We calculate that our model’s α in

contemporary Mexico as .233.Fourth, the theory predicts that SNGs which face electoral

competition have a stronger preference for providing public goods over rents. We provide

evidence of good governance in municipalities that have become electorally competitive.

Finally with respect to good governance and economic growth, we argue that increases in

α and in electoral competition lead to better governance. We present evidence of the change in

the rates of growth in Mexico as α and electoral competitiveness change.

Our approach contrasts with two normative rationales for federal revenue sharing. First,

the traditional normative public finance perspective, relying on the assumption of a benevolent

government that maximizes citizen welfare, argues that the federal revenue sharing allows the

government to redistribute income, which a decentralized set of SNGs could not achieve on their

own. Second, Persson and Tabellini (1993) argue that revenue sharing in federal systems

provides a degree of social insurance against regional shocks. From a normative perspective,

neither approach can be criticized. From a positive perspective, particularly in the developing

world, these rationales often fail. Mexico’s widespread corruption, its many anti-development

policies, and the long-term diversion of public resources to maintain the party in power all

question relevance of the benevolent government assumption. As we will show, Mexico’s initial

revenue sharing scheme involved little redistribution and no social insurance: states received

back from the center in proportion to their contributions. Although the Mexican system now

contains some redistribution, it is a small portion of the total revenue shared with the states. Put

7

simply, the normative rationales cannot explain why Mexico pursued fiscal centralization and

revenue sharing.

An alternative hypothesis, more consistent with the facts in many developing countries, is

twofold. First, revenue sharing qua FPWD allows the federal government to achieve a monopoly

on taxes. The tax monopoly allows them to extract more revenue from citizens than can the

SNGs in the aggregate. Capital and labor mobility, even if imperfect, limits the ability of SNGs

to extract taxes from their citizens and firms. The federal government, with its monopoly on

taxes, does not face this limit. Second, Mexican politicians used these funds to keep their

dominant party in power for decades. Diaz-Cayeros, Magaloni, and Weingast (2000), for

example, show that the dominant party used its control over fiscal resources to punish those areas

that elected the opposition by withdrawing funds

Our approach has implications for the larger question about the political foundations of

good government. Appropriately structured federalism helps improve the incentives for resource

allocation, particularly when corruption throughout and bad central government economic

policies represent major problems hindering economic performance. Appropriately structured

federalism decentralizes a range of decisions, taking a series of allocation decisions away from

the federal government. This can have efficiency gains when the federal government is a major

source of inefficient policy (Jin, Qian, and Weingast 1999; see also Brennan and Buchanan

1980). The key to capturing efficiency gains through decentralization, however, is getting the

incentives for local government officials right.

The paper proceeds as follows. Section 2 develops our theoretical framework. Sections 3

and 4 derive, respectively, the effects that revenue sharing and political competition have over

good governance. Section 5 provides the background on federalism and electoral

8

competitiveness in Mexico in Mexico. Section 6 provides evidence supporting our approach

from the Mexican fiscal system, including the application of the fiscal law of 1/n, and provides

evidence to support the theory’s predictions. Our conclusions follow.

2. A Model of SNG Choice under Revenue Sharing

We begin by analyzing how fiscal incentives affect a subnational government’s (SNG) decision

making. We assume that government officials, whether elected or not, seek to advance their

careers by remaining in office or, if they are part of a strong national party system, by enhancing

their party‘s local reputation. Government officials pursue these career goals by attempting to

generate political support. Democratically elected representatives must maintain the support of a

majority of voters. For authoritarians, the percentage required to retain their office is

undoubtedly lower than a majority, though it is also substantial portion of the population.5

We assume that government officials have two separate means of fostering political

support. First, they may engage in corruption and provide politically generated rents to

constituents. These rents may take a variety of forms, such as corrupt payments or various

benefits derived from market intervention (e.g., local monopolies or anti-competitive

regulations). Second, government officials may provide public goods that foster local economic

growth, providing indirect benefits to constituents.

5 Students of comparative politics again and again emphasize the need of authoritarian governments to maintain sufficient support to remain in power. See for example Geddes 1999, Przeworski and Limongy 1997, Poole and Londregan 1996.

9

In what follows, we study how various fiscal arrangements affect political officials’

tradeoff between pursuing rents and providing public goods. Assume that each of N

(subnational) governments seeks to maximize the following utility function,

),,( yrUU ii = (1)

where r represents the level of corruption and rents and y is the level of the (market-fostering)

public good. The utility function is that of a representative constituent. For democratic local

governments, the utility function is that of the median voter. By maximizing the median voter’s

utility function, the SNG assures enough support to survive. As noted, hegemonic SNGs also

need to retain political support, though less support than those which face electoral competition,

so we assume that these governments also have a representative constituent.

For simplicity, we drop the subscript i. We assume that U is increasing but at a

decreasing rate in each variable; that is, U . We will also assume

that U , indicating that the two means of providing political support do not interfere with

one another, if at all, they actually help one another.

),( yr

,0 <yyU 0,0,0 ><> yrrr UU

0≥ry

We also assume that the SNG faces a budget constraint. Its budget, B, is given by

, where T = T(θ,τ(y)) represents the amount of fiscal transfers from the national

government; τ(y) is the locally generated taxes as a function of the level of public good provided;

and α is the proportion of locally generated taxes that the SNG may keep ( ). The public

good, y, is assumed to affect positively the locally generated taxes; that is, τ

)(yTB ατ+=

10 ≤≤α

y > 0.6

Finally, we have the SNG’s budget constraint, which allows the SNG to spend its budget

on r and y:

10

r (2) Bay ≤+

where a is the parameter representing the relative cost of a unit of y in terms of r (we normalize

the price of r to 1). Substituting for B and rearranging terms yields . ayyTr −+= )(ατ

To study the government’s behavior, we use expression (2) for r and substitute back into

(1). This yields the following maximization problem:

),,)(( yayyTUMax y −+ατ (3)

with first order conditions:

.0)'( =+− yr UaU ατ (4)

The first order conditions imply that the government’s marginal rate of substitution between r

and y, U , is given by: yr U/

)'(1

aUU

y

r

−=

ατ. (5)

To interpret the first order conditions, suppose that y had no effect on taxes; i.e., . Then we

would obtain the standard result that at the optimum choice of r and y the marginal rate of

substitution between S

0'=τ

r and Sy equals minus the ratio of the cost of r to the cost of y; that is,

aUU

y

r 1−= . In our problem, y affects the taxes, and hence the term in equation (5), ατ’, which

adjusts the marginal rate of substitution. Intuitively, the adjustment term reflects y’s positive

effect on the budget, which leads the SNG to choose more y than otherwise. The specific amount

of the adjustment in the marginal rate of substitution is given by ατ'.

6 Perhaps a more accurate formulation follows Haughwout and Inman (2000a,b) and lets the size of the local economy, E(y,r,τ) be positively related to y and negatively related to r and τ, so that locally raised revenue is τE(y,r,τ). As the results are qualitatively similar, we use the simpler formulation.

11

To assure we have a maximum, we investigate the second order conditions. Under the

assumption that , these conditions hold. This ensures that an equilibrium exists and

is characterized by a point.

0/2 >∂∂∂ yrS

We now turn to our principal question about how the SNG’s behavior changes in

response to the fiscal system. To begin, we investigate how a shift in α affects the SNG’s choice

of r and y. To do this, we derive the comparative statics results of α on (r*,y*), the SNG’s

optimal choice of r and y. We start with y*. The first order conditions (4) give y* as an implicit

function of the exogenous parameters (a, α, t). By the implicit function theorem, we have for any

exogenous parameter, xi, that:

yFxF

xy i

i

∂∂∂∂

−=∂∂ , (6)

where F is the left hand side of the first order conditions (4).

To simplify, we know from the second order conditions that the denominator of (6) is

negative for all xi. Thus, to sign ( ), we have to sign the numerator; that is, sign (

= sign ( . To calculate the effect of α on y*, we differentiate (4) with respect to α.

Given our assumptions, is positive. So increasing α, the SNG’s share of locally generated taxes

has the effect of increasing the SNG’s optimal level of y*.

ixy ∂∂ / )/ ixy ∂∂

)/ ixF ∂∂

The effect of α on r* is more complicated. To calculate this, we first observe that the

budget constraint yields . Totally differentiating this equation and setting

the other exogenous changes to 0 yields the following:

** )),(,( ayytBr −= ατ

.

**

)( ατατ

α tyt ByaBr+

∂∂

−=∂∂ (7)

12

Equation (7) has an interesting interpretation. First, since the first term on the right hand side is

negative and the second term is positive, ∂ can either be positive or negative. Second, the

response of r* to an increase in α is negatively related to the∂ . That is, the larger is the

effect of α on y*, the smaller is .

α∂/*r

α∂/*y

α∂∂ /*r

In sum, increasing α, the proportion of locally generated taxes retained by the SNG,

increases the resources the SNG devotes to the market-fostering public good, y. The reason is

straightforward: producing y has two benefits. As a direct benefit, more public goods make

constituents better off and thus increase constituency support. As an indirect benefit, more y

produces a healthier economy, thus more budget revenue, allowing the SNG to increase both r

and y. The larger is α, the greater the SNG’s provision of the public goods and the lower is its

reliance on rents to maintain political support.

We illustrate this effect in figure 2.1 which shows how τ(y) affects an SNG’s choice of r

and y. To make the case simple, we let the parameter a, giving the cost of y relative to r, equal

one. When α = 0, indicating that the SNG captures none of the increase in locally generated

taxes, τ(y), the SNG’s budget line is then given by the straight line in figure 2.1, . In

this case, the SNG choose the combination (r

Tyr ≤+

o,yo).

When the SNG captures a portion α > 0 of τ(y), the budget curve shifts outward in a non-

linear fashion, as indicated on the figure. The shift makes the SNG’s budget non-linear because

the budget increase, ατ(y), increases as the SNG chooses more y. When y = 0, r = T and the

budget shift is zero; when r = 0, y = T and the budget shift is largest.

This example shows how the effect of y on its budget induces the SNG to substitute r for

y. In the figure, r decreases somewhat from r0 to r1 while y increases from y0 to y1. Per our

13

comparative statics result, the effect of ατ(y) on y is larger than the effect on r. This reflects the

fact that y not only increases S directly but shifts out the budget line.

Figure 2.1

3. Applying the Model: The Economic and Political Effects of Revenue Sharing

Federal systems around the world employ a range of revenue sharing schemes. To investigate the

incentive effects of revenue sharing, we compare two tax and revenue schemes. In the first, the

14

SNG raises all its own money, so that T = T0 and α = 1. In the second, the national government

captures all locally generated revenue, possibly combining this with other revenue, and then

divides the pool among all the SNGs. In the simplest scheme, the national government collects

all the revenue from all SNGs and then divides this revenue among n SNGs so that each receives

1/n of the total common revenue pool. Because the government takes all locally generated

revenue from all SNGs and then returns 1/n of the total to each, α = 1/n.

The Fiscal Law of 1/n and FPWD

We compare the effects of these two fiscal systems using equation (5), )'(

1aU

Uy

r

−=

ατ.

The term in the (right hand side) denominator tells us, in comparison with the case where y has

no effect on the SNG’s revenue, how much the SNG’s choice of y shifts the marginal rate of

substitution at the optimum toward y for a given α > 0. When the SNG keeps all locally

generated revenue, α = 1, so the shift in the marginal rate of substitution is the full τ'. Next we

calculate the shift when the SNG keeps none of its locally generate revenue and receives all its

revenue from the common pool, P, created by the revenue from all SNGs. In this case, the SNG’s

budget is given by , where )(*0 yTB τ+=ny

nyP )()]([ ττ

+−

=T . The last equation says that

the total transfer, T, can be divided into two portions: first, the SNG’s share of the revenue raised

in all other SNGs; and second, its share of the tax revenue raised in its own jurisdiction, τ(y)/n.

This implies that, when revenue from all SNGs goes into a common pool P, α = 1/n. This yields

the shift is in the marginal rate of substitution as τ'/n.

From this analysis we derive the fiscal law of 1/n: When the SNG raises all of its taxes, it

chooses more y because it captures all the subsequent increase in revenue due to τ′; In contrast,

15

under revenue sharing, it captures only τ′/n. In short, the relative effect of an increase in y* under

complete revenue sharing scheme is 1/n the magnitude of the increase when the local

governments captures all the locally generated revenue. Under revenue sharing scheme, the SNG

must share nearly all the revenue increases (i.e., (n-1)/n of the increase) with the other SNGs,

while it retains only 1/n of the revenue increase. In contrast, when an SNG captures all its

locally generated taxes, however, it keeps 100 percent of any increase in taxes from increasing y.

For federal systems with many SNGs -- Mexico with 32 states, Russia with 89 regions, or the

United States with 50 states – the disincentive to invest in public goods can be large.

To illustrate the nature of the shift, we calculate an example. We let the relative cost

parameter, a = 1, the number of SNGs be 32, and . If there were no shift due to

τ(y), i.e., if α = 0, then the marginal rate of substitution between r and y at the optimum (r*,y*) is

given by U . Under the fiscal arrangement where the SNG gets to keep all its revenue,

α = 1, the shift in the marginal rate of substitution is given by ατ′, which equals 1/4 in this

example. This implies that the SNG’s marginal rate of substitution at the optimum is:

4/),,( yryE =ττ

1/ −=yr U

34

4/111

=−

=y

r

UU .

In this case, the shift in the marginal rate of substitution produced by the effect of y on local

taxes accruing to the SNG is a full 1/3.

Under the fiscal arrangement where the SNG keeps none of its revenue, but the federal

government shares all collected SNG revenue equally among n SNGs, α = 1/n. Given n = 32,

then the SNG’s marginal rate of substitution at the optimum is:

127128

)/1(*)4/11(1

=−

=nU

Uy

r .

16

In the common pool case, the shift toward y induced by τ(y) is just 1/127.

To summarize the effect in this example, consider the base case where y has no effect on

the budget. Then the shift marginal rate of substitution away from the base case when α = 1 is a

full 1/3 and is more than 42 times the shift when α = 1/n = 1/32.

Many parts of the world during the mid to late twentieth century exhibit a trend toward

greater fiscal centralization. In Argentina and Mexico, for example, the national government

made a political deal with its SNGs in which the SNGs gave up important aspect of their taxing

authority in exchange for the national government providing greater revenue. National

governments were able to offer this exchange in part because they gained monopoly authority

over taxes. As Hayek (1939) and Tiebout (1956) noted long ago, competition among SNGs for

tax base limits their ability to tax: too high a tax rate induces mobile factors to move to other

jurisdictions. The national government faces no such constraint because it can impose the same

tax laws across all SNG jurisdictions. Thus, when the SNGs give up their tax authority to the

national government, the national government can set taxes at higher rates than allowed by SNGs

facing competitive pressures.

We call this agreements fiscal pacts with the devil for two reasons. First, revenue sharing

agreements typically require that SNGs give up their policy and fiscal autonomy. Second, these

pacts induce the SNG to substitute non-productive transfers, such as corruption and rent seeking,

for public goods. The tradeoff between r and y represents SNG’s tradeoff between corruption,

rent-seeking and market intervention, on the one hand, and economically productive activities on

the other. The simple model above shows that, holding constant for all other proclivities toward

corruption and market intervention, the fiscal system affects an SNG’s incentives to foster

markets. Put simply, the stronger an SNG’s fiscal incentives –as measured in terms of the

17

proportion of locally generated taxes it captures –the less corrupt and more pro-market oriented

the government. All this generated simply by government officials pursuing their rational goal of

political survival.

The model predicts that countries whose fiscal system primarily finances SNGs through

revenue sharing should have higher rates of corruption than countries whose SNGs are primarily

self-financing.

4. Applying the Model Part 2: The Effects of Political Competition on Good Governance

We now investigate how electoral competition systematically affects local leaders’ preferences.

As noted, all SNGs need some level of political support in order to survive. Maintaining

political support in localities where the government faces political competition requires greater

attention to public goods provision than in areas where the government retains monopolistic

control of elections. This is because an SNG that faces competition must appeal to the median

voter, while a hegemonic SNG does not need to.

We model the difference between competitive and hegemonic SNGs as a shift in the

representative voter’s utility function that they must satisfy. As shown in figure 4.1, the utility

function for localities without competition, U(r,y;0), is systematically lower (prefer more r,

ceteris paribus) than the representative utility function for localities facing competition, U(r,y;1).

Along the flat budget line, α = 0, the two types of leaders make different choices between r and

y, with the SNG facing greater competition choosing more y and less r than the leader who does

not face competition.

18

To study how the fiscal incentives work differently in these two localities, we examine

the shift in SNG choice when we increase α a fixed amount from α = 0. Preferences of SNGs

without competition are in the lower right. In this region, the fiscal leverage associated with an

increase in α from α = 0. In the figure, this SNG provides a tiny increase in resources devoted to

corruption and a slightly larger amount to public goods.

In contrast, the preferences of leaders from a locality with a high degree of competition

are in the upper left. In this region, a shift in α affords the SNG significant leverage, and it uses

the increase in fiscal resources to provide approximately forty percent more public goods. There

is also a small increase in corruption and rent seeking.

Figure 4.1

19

In sum, political competition makes SNG leaders more sensitive to constituent interests. The

model predicts that political competition combines with the fiscal effect to give local leaders

greater incentives to provide public goods.

5. Historical Background on Fiscal Federalism and Electoral Competitiveness in Mexico

We present here the general trends in twentieth century Mexico with respect to the fiscal

federalism and electoral competitiveness.

A. Fiscal Federalism

Although the Mexican Constitution of 1917 established a decentralized fiscal system,

revenue collection and expenditure became very centralized over the second half of the 20th

century. We divide the evolution of the Mexican fiscal system into four periods. The first period

covers the end of the Revolution to the late 1940s. States retained full fiscal authority, but the

center was too weak to police the common market, and the system suffered from various

common pool problems. The second period covers 1947 until end of the 1970s. The federal

government slowly centralized fiscal power by coaxing all the states to enter a revenue sharing

pact. The third period, covering 1980 to 1994, began with the establishment of a formula to

allocate federal revenue among the states, all the states were in the FPWD and it was a period of

great centralization. Finally, the fourth period covers from 1994 to the present. We discuss these

periods in turn.

20

Period 1: 1917-1947. The victors of the Revolution (1910-1917) drafted a constitution

with the aim of restoring political order and financial stability. The constitution granted exclusive

taxing powers over strategic areas to the federal government and forbid the states from levying

taxes that would obstruct the emergence of a national market. Yet the constitution failed to

achieve these objectives. Well into the 1940s, the fiscal system remained chaotically

decentralized as the Mexican federal system exhibited significant common pool problems. The

states levied unconstitutional and overlapping taxes; they also erected internal trade barriers; and

the federal government proved unable to police the constitution or the common market (Margain,

1971; Islas 1997, Diaz Cayeros 1999). To illustrate, in 1940 the three levels of government

levied 80 separate taxes on industry and commerce and 32 on capital (Presidencia de la

Republica, 1997). Exacerbating these problems, “corruption was rampant in every tax

administration” (Islas 1997). The result was too many taxes, protectionism that hindered

interstate commerce, and corruption.7

In the 1920s and 1930s leaders and interests in many states realized that there existed

efficiency gains to improving cooperation among the states (Diaz Cayeros, 1999). However,

consistent with the two dilemmas of federalism (de Figueiredo and Weingast 2000), the states

failed to cooperate in the construction of a national market for two reasons. First, it was hard for

states to trust one another because each had incentives to defect, and most states retained the

necessary resources to re-start a war. This is evidenced by the smoldering civil war that persisted

well into the 1930s. Second, it was hard for states to trust the federal government as a neutral

enforcer of the national market (Diaz Cayeros, 1999).

7 According to Diaz Cayeros (1999, 48), “The constitution prohibited taxes that would hinder interstate trade, but this was systematically violated. Local governments relied on immobile assets or on transactions that were easily measured. Producers had few exit options due to the state of physical infrastructure and the financial system. This state of anarchy hindered economic growth but politicians were unable to solve the collective dilemma.”

21

Period 2: 1947-1970s. Two contradictory forces were at work during the second period.

First, in the initial phase of centralization, the federal government reduced the ability of the states

to overgraze the commons. The political result was a federal system closer to market-preserving

federalism (Montinola, Qian, and Weingast 1995, Weingast 1995). The economic result was

greater economic efficiency, as witnessed by sustained growth commonly known as the

“Mexican Miracle.”

However, by coaxing state after state to join the FPWD after 1947, the federal

government progressively compromised the states’ ability to exercise independent policy and

fiscal autonomy. By the end of the second period, fiscal pacts with the devil had significant

political effects beyond their direct economic effects. By agreeing to these pacts, the states

abdicated their right to tax several important areas of economic activity in exchange for revenue

transfers and investment projects from the federal government. Also, the states would cease to

collect unpopular local taxes.8 By centralizing tax collection and expenditure, the federal

government increased its revenue and control over policymaking.

The federal government induced the states to join a FPWD through economic and

political carrots and sticks. First, the federal government offered the carrots of additional revenue

for states and exciting political careers to the state officials. Second, the federal government

provided two sorts of sticks: first, it would collect federal taxes in the state used for the revenue

sharing pool regardless of whether the state joined the pact; and second, the PRI’s system of

rotation for political officials destroyed the ability of representatives to develop independent

relationships with their constituents.

8 According to Islas (1997), federal officers argue that states lack the political courage to raise their own taxes. “Appointments of tax collectors are often politically motivated and those selected are often not committed to the task or are strongly influenced by political pressures, such that relatives of the local authorities or powerful “caciques” do not pay their taxes at all” (Islas, 1997)

22

In 1947 the federal government created a new, voluntary and contractual tax coordination

system. States that joined would get more revenue in exchange for their authority to tax in the

areas decreed exclusively federal. States that did not join could continue levying taxes in these

areas, but they would continue to bear the cost of the federal taxes without receiving any

revenue.

Between 1947 and 1952, eleven states joined this system: four under the control of the

federal government: the Federal District and the territories of Quintana Roo and North Baja

California and South Baja California; and seven voluntarily: Aguascalientes in 1949; Morelos,

Queretaro and Tlaxcala in1950; Michoacan and Sinaloa in1951; and San Luis Potosi in 1952.

The next step in the fiscal centralization occurred in 1953 with the first Law of Fiscal

Coordination (LFC). As with the previous agreement, the LFC was a contractual arrangement

between two levels of government. The LFC’s intent was to coax the still reticent states to join

the revenue sharing system. This law also declared the income tax exclusively federal; and the

center committed to share a proportion of it with the states. This carrot was sufficiently attractive

to induce seven more states to join: Colima, Yucatan, Hidalgo, Campeche, Tabasco, Puebla and

Guerrero. Nonetheless, for the next two decades, fourteen states remained outside the system

In 1972, the federal government increased the sales tax by one percent, sharing the

increase with the states. The additional revenue induced even the most reticent states to comply

with the voluntary tax coordination system (Presidencia de la Republica,1997: 88).

The federal government completed its program of fiscal centralization in 1980, when it

issued a new LFC. In exchange for greater centralization of tax authority, states received

increased and supposedly less arbitrary transfers that would be regulated by a formula.9 Prior to

9 The new law also introduced two big changes in the revenue sharing system. First, it created the value-added tax (impuesto al valor agregado, IVA), replacing a vast number of municipal and state taxes on production and services.

23

1980 the central government shared with the states a portion of two specific taxes (the sales tax,

and income tax); since then it has shared a proportion of a larger set of taxes. The 1980 law also

markedly increased the proportion of revenue that the central government transferred to the

states.

Period 3: 1980-1994. The third period of Mexican federalism starts with the 1980 reforms.

This new system completed the centralization of tax authority and established a formula to

provide transparency to the assignation of the Transfers fund. Stability was not established,

however, as the central government altered the formula frequently: in 1981, 1983, 1984, 1988

and 1990.10

Period 4. 1994-Present. Since 1994, population and revenue raised in a state are accorded

equal weight (45.17% each) in the distribution of funds. We analyze this formula in section 6.

Moreover, the fiscal constraints on the national government have allowed some freedom to

emerge for lower governments, particularly municipalities.

B. Electoral competition

Between its formation and the late 1980s, the PRI controlled nearly all elected and

appointed offices at all levels of government. This began to change after the fraud ridden and

highly contested presidential elections of 1988.11 In 1988, opposition parties governed only 39 of

the 2,419 municipalities, (1.6% of the population); by May 1999, these parties governed 583

municipalities (46% of the population). In 1988 all 32 states had a PRI governor; by May 1999

Second, the law established the National System of Fiscal Coordination (SNCF). Under SNCF the federal government agreed to share with the states and municipalities a portion of the Recaudacion Federal Participable (RFP).. 10 We analyze two of these formulas in section 6; Appendix 1 summarizes the various changes in the formula. 11 The following data comes from Lujambio 2000.

24

10 states (32% of the population) had opposition governors, including the Federal District.

Similarly, in 1988 the PRI had the necessary supra-majorities to legislate in all matters in all the

nation’s legislative assemblies; by May of 1999, the PRI had this level of control in only two of

the 34 assemblies. The opposition’s success deepened in the 2000 presidential elections where

for the first time a party other than the PRI won the presidency.

These changes were brought about by different political and economic developments: the

fiscal changes discussed above in combination with the international debt crisis and the fall in oil

prices in the early 1980s, the breakdown of the elite coalition in the PRI and a split of the party

in 1986, several electoral reforms of which the last (1996) was approved by all the parties

represented in congress. This is not the place to analyze the sources of these changes, but for the

purposes of this paper, after 1988 electoral competition had become a reality in many

municipalities and states.

6. Empirical Evidence about Mexico’s Fiscal Pacts with the Devil

In this section, we provide five pieces of evidence from Mexico to support the predictions of our

model: we account for why different states joined the FPWD at different times; we provide

evidence about changes in the spending and taxing behavior of states after entering the FPWD;

we calculate α from the different revenue sharing formulas used after 1980; we provide evidence

of the changes in good governance with the emergence of political competition; and we support

our hypothesis of the of changes in the rate of growth as α changed.

25

A. Explaining which States Delayed Joining the FPWD

Given that all Mexican states faced the same carrots and sticks for joining the FPWD,

what explains the differential willingness to enter these agreements? Fourteen states, the Federal

District and two territories (these latter three governed directly by the federal government) joined

in the late 1940s and early 50s. In contrast, fourteen states waited over two decades to join, in

1972. Our theory predicts that the more market-oriented states are likely to be the last to join.

The reason is that, ceteris paribus, the marginal effect of public goods on local tax revenues

differs across states.

To test our hypothesis, we investigate the relationship between when states joined the

FPWD and various economic indicators. The analysis reveals that, per our predictions, the richer

and more market oriented states on average joined later. For example, the average per capita

state income of states joining in the 1940s and 50s was approximately 5,600 pesos in 1970;

whereas for the states that joined in 1972, it was approximately 16,000 pesos, roughly three

times higher. Similarly, states that joined later had far more exports. At present, we only have

export figures for 1998, so this figure is only indicative. States that joined the FPWD early

averaged exports of $1.2 billion in 1998, whereas those joining late averaged exports of $4.4

billion, a factor of roughly 3.7 times larger.

To obtain more systematic evidence, we used logit analysis. To create the dependent

variable, we divided states into those who joined early and late. We used two independent

variables, state GDP per capita in 1970 and state exports in 1998.

The results are reported in Table 6.1. They show that GDP per capita is a statistically

significant determinant of the decision to join early. In contrast, the variable, exports in 1998, is

not statistically significant (perhaps because our data for this variable is for a date so much later

26

than when these states made their decisions to join). The overall performance of the logit is also

good. For example, it correctly predicts 74.2 percent of the cases, which out performs the null

model’s correctly predicting 54.8 percent of the cases.

The evidence supports our theory. On average, states with higher per capita income

joined later.

B. The Impact of the FPWD on Governance

We have argued that, by breaking the relationship between the provision of local public

goods (more broadly, good government) and state revenue, FPWD diminished state officials’

incentives to foster a prosperous economy. After joining the FPWD, states should thus substitute

away from public goods to non-productive expenditures, such as corruption.

Table 6.1: Logit Analysis of When States Joined the FPWD Independent Variable Estimated Coefficient (t statistic) Constant -2.36* (2.47) GDP/capita 0.25* (2.02) Exports -.075 (.44) No. observations 31 Percent Correctly Predicted

74.2

Log Likelihood Initial -21.5 Convergence -15.2 * Statistically significant at the .05 level.

27

Table 6.2 provides some modest evidence of an increase in rents over public goods after

the states all joined the FPWD. The table shows the percentage growth of employment in the

government and in the economy. In the period 1910-1970, public sector employment was 1.2

percent of total employment; by 1970-1983 it had become 20 percent. Consistent with our theory

about subnational governments under the FPWD, public employment in municipalities and states

grew even faster than federal employment, and at a time when policy responsibly shifted to the

federal government. Although some of the public sector growth represents an increase in public

services, some of it reflects politicians offering jobs to supporters as part of the PRI’s extensive

patronage system. Indeed, aviadores are common throughout the public sector in Mexico: people

on the payroll who never show up to work but receive their checks.

6.2 Rate of growth of Employment in the Public Sector compared with the Economy 1910-1970 1970-1975 1975-1983 1970-1983 % Growth of total employment

243

118

127

151

% Growth of govt. employment

958

348

185

647

% Growth of Federal govt. Employment

1142

350

182

639

% Growth of Municipal and State govt. employment

505

339

203

690

Ratio of Public sector /total employment

1.2

4.8

14

20.4

Source: the authors with data from Zaid 2000

28

Our theory also makes several predictions about tax collection following the FPWD.

First, it predicts that state taxes should go down after accepting a FPWD. As α decreases, states

have incentives to substitute towards non-productive expenditures; as shown in the section 3.B,

tax collection effort should also fall, so locally raised revenue should fall. Second, we argue that

the federal government created the system in part to capture a central monopoly on taxes,

allowing it to extract greater resources from the economy. In other words, our theory predicts

that total federal taxes should rise following the FPWD. We discuss these predictions in turn.

Ideally, we could use time series data on state taxes to investigate how taxes fared before

and after each state joined the FPWD. Such data is difficult to obtain. Instead, we look at the

behavior of all states following the 1980 changes. As suggested above, this date represents a

watershed year in the fiscal system, as state dependence for revenue on the federal government

increased dramatically.

The effect of the 1980 changes on state taxes is easily seen in figure 6.1. In the decade

prior to 1980, state taxes averaged a little above one percent of GDP. Over the next decade, state

taxes fell continuously so that, by the end of the decade, they were below one-half a percent of

GDP. 12 The figure also shows the incentive of “bribe” aspect of the revenue system: the fall in

locally generated taxes is more than compensated by the rise in revenue sharing.

12 A recent econometric study, for example, claims that after 1980, for every peso given through transfers, state treasuries reduced their own tax revenue by 17 cents, (Gutierrez and Islas, 1995; 139).

29

Figure 6.1. State Revenues as a Percentage of GDP

0

0.5

1

1.5

2

2.5

1970

1971

1972

1973

1974

1975

1976

1977

1978

1979

1980

1981

1982

1983

1984

1985

1986

1987

1988

1989

1990

1991

1992

Years

Rev

enue

s/G

DP

Ow n sources

Gov. Transfers

Source: Islas (1997)

Another way to measure the fiscal impact is to assess the proportion of revenue raised by

a state from its own taxes. The ideal data set would cover the entire period, so we could see the

differential impact of the early FPWD on both types of states (early and late joiners).

Unfortunately, we have data only from 1970 onwards.

The data reveal significant changes following both the 1972 and 1980 centralizations. Per

our theory’s predictions, state taxes as a proportion of total state revenue fell dramatically after

1972: from 45 percent prior to the 1972 to 29 percent for the rest of decade (table 6.3). This

proportion fell again dramatically after 1980, to 3 percent, remaining slightly lower in the late

1990s.

30

Table 6.3: State Taxes as a Proportion Of Total Revenue (percent)

1970-72 45 1973-80 29 1981-94 3 1994-99 2

Source: INEGI

We now turn to our second prediction – that federal taxes exhibit a greater rise than the

decrease in state taxes. Systematic data for all the different federal taxes is hard to obtain.

Nonetheless several indicators are consistent with our theory. First, consider the change in the

federal taxes after the fiscal changes in 1980 (see figure 6.1). Federal taxes in 1980 were

approximately 15-16 percent of GDP, with oil revenue representing approximately half of the

total. The 1980 law imposed a new value added tax, generating new revenues of between two

and three percent of GNP. This represents an increase in non-oil revenue on the order of one

quarter to one third. Similarly, collection of the income tax (created in 1972, the year of the final

FPWD) increased 22 percent every year since its creation.

A final illustration concerns the federal sales tax. When states joined the FPWD in 1972,

the sales tax became exclusively federal, and the states outside this agreement stopped collecting

this tax. At this time, the federal government also increased the rate of the federal sales tax to

four percent. Even so, collection of the sales tax increased far more than was projected based on

the conditions of the previous presidential administration (see table 6.4). Four years later, in

1976, actual receipts were over two and one half times anticipated receipts.

31

Table 6.4 Effects of FPWD on the Collection of the Federal

Sales Tax (million pesos) Year Projected

collection* Total collection Unanticipated

increase (%)** 1973 6.0 7.6 26 1974 6.8 10.6 55 1975 10.2 17.4 95 1976 11.7 28.7 170

Source Escamilla de León (1999) * projection based on the prevailing tendency in the previous presidential term ** between projected and actual

In short, the evidence supports both of our theory’s predictions about the changes in tax

revenue.

C. Calculating “α” and the Incentive Effects in Mexico’s Revenue Sharing Formula

We now investigate incentive effects of the various formula employed in Mexico’s revenue

sharing system after 1980. In particular, we calculate the marginal retention rate, α, for three of the

formula employed over the period.

(1) Analysis of the 1980 formula. We begin with the formula used in 1980,

∑

=j

i AAiR

where Ri is the revenue share allocated by the formula to state I from the common pool, P, where

, and A∑=j

jAP j is the amount of revenue raised in state j. The formula says that a state’s revenue

share, Ri, is the fraction of the total revenue pool represented by the share of taxes raised in its state.

Our theory yields several implications about this revenue sharing formula. First, the formula

provides good incentives on the portion of revenue allocated by formula: a state captures 100

percent of any increase in local revenue flowing into the pool. Second, because only a small portion

32

of funds spent in localities is allocated by this formula, the incentive effects of the centralized fiscal

system are actually much worse. In 1980, the amount of funds spent by the federal government in

the states on local projects was approximately four times that allocated by the revenue pool. Thus,

although the funds allocated by formula are on the order of only one fifth of those spent locally,

implying that α is on the order of .2. Third, the negative effect on locally generated taxes is likely to

be larger in the more market oriented states. The reason is that in these states the effect of public

goods on locally generated revenue is likely to be larger in these states.

The 1980 formula involves no redistribution to poorer states. Remarkably, each state puts in

Ai and receives back Ai. To see this, notice that state i’s share of the total pool, P,

is PA

ARiP

jj

i )(∑

= . Substituting for P, yields that each state gets back Ai.

These results for the revenue sharing formula in used in 1980 demonstrate that the purpose

of this FPWD and under this revenue scheme cannot be redistribution. Because each state gets back

the same amount that it puts in, the purpose of revenue sharing must be political and not

redistributive.

(2) Analysis of current formula (1995-present). Mexico’s current formula for distributing

revenue dates from 1995 and has three components. The formula allocates 45.17% of the revenue

pool on the basis of population; 45.17% by a formula, analyzed below; and 9.66% in inverse

proportion to the other two criteria.

The component based on population is simple to analyze. A state with a proportion, qi, of

the population receives .4517qi of the total pool. The formula entitles a state of average population

(3.125% of the total) to receive .4517*.03125P = .014P. This also implies that, if a state changes its

33

policies so that it increases its revenue by x%, this component entitles the average state to .014x, or

about 1.4% of the increase.

We analyze the 45.17% of total revenue allocated under by formula. It implies several

different effects. After 1980 the federal government attempted to give states an incentive to improve

their collection of tax revenue. The 1995 formula does this in a clever if complicated way.

Since 1995, Mexico has used the following formula (F1) for distributing 45.17% of the pool

based on revenue collection:

∑ −

−+ =

jjtjtjt

itititit AAR

AARR

)/()/(

1

11 (F1)

where Rit is state i’s share of the total revenue in time t; Ait is the proportion of revenue pool, P,

raised in state i in time t; and the total revenue pool, P, is given by . ∑=j

jtAP

The incentive effects of this portion of the formula are good: the formula returns any

increase in tax revenue contributed to the common pool. Mathematically, this is easily shown since

Intuitively, this can be seen by considering what happens when one state raises its tax

collections by x%, assuming that the revenue from all other states remains constant. The numerator

of the formula becomes , and the denominator becomes . Suppose that

state i produces 5 percent of total revenue and that it increases collections by 10 percent (i.e., x = .1).

Then the numerator becomes 1.1*.05 = .055; and the denominator becomes .95 + .1*.05 = 1. So a

ten percent increase in locally generated revenue leads to a ten percent increase allocated by the

formula. To summarize, the formula says that, holding constant for the behavior of other states, if

state i increases its revenue, this year over last year, then next year it will receive nearly the full

increase. Of course, this formula applies only to 45.17% of the total pool.

.1/ =∂∂ itit AR

itRx *)1( + ∑ +j

itjt RxR *

34

We now investigate the impact of the formula as a whole on a state’s marginal incentives.

Thus, if state i increases its collections by x%, the overall formula has three independent

components that affect state i‘s portion of the increase. First, by the population component, state i

with population proportion receives back a portion of their gain from the common revenue pool

of .4517q

iq

i*x. For a state with the average population of .03125, this implies that state i receives

back 1.4% of the gain. The second component is by the complex formula, F1, which grants state i

45.17% of the gain. The third component, based on inverse of population, grants a state with the

average population an additional 0.3% x.

Thus, for a state with the average population, an increase of x% of revenue translates into

three components of gain are: .4517x, .014x, and .003x, for a total increase of .466x. Of course,

the overall formula represents only half of all federal revenue spent in the states. This implies

that, at the margin, each state keeps a little less than one quarter (.233) of any increase in locally

generated revenue.

(3) Analysis of the 1991 formula. As indicated in the Appendix, the revenue sharing

formula evolved in the early 1990s. The difference between these formulas and that initiated in

1994 is solely the weights accorded the three portions allocating funds based on population,

revenue collection, and the inverse of the first two. As shown in table A.1, the weights changed

in each year from 1991-94.

The analysis of the 1994 formula can readily be applied to these earlier formulas. The

1991 formula is perhaps the most interesting of the group. This formula allowed states to capture

72.29 percent of their revenue collection. Recalling that the revenue allocated by formula is

about half that spent by the federal government in the states at this time, the 1991 formula

implied a higher α than the 1994 formula, on the order of .37. As table A.1 reveals, the

35

proportion allocated by on revenue collection systematically decreased from 1991 to the present

formula analyzed above so that α decreased from approximated .37 in 1991 to .23 in 1994.

Implications of the revenue sharing formula. Over the 1980s and early 1990s, Mexico

changed its formula for distributing revenue from the common pool. Throughout this period, the

system provided poor incentives for improving tax collection and promoting local growth.

Unfortunately (from the incentive point of view), the revenue going into the pool and thus

covered by the three-part formula just analyzed is only a portion of total funds spent by the central

government in the states and localities. The amount of discretionary money given states under other

categories — earmarked funds referred to as federal investment -- far exceeds that given in this

category. In 1982, federal investment in the states on average exceeded transfers by a ratio of four to

one. In the late 1990s, Federal investment approximately equaled transfers. Further, due to Mexico’s

centralization, including the FPWD, states have many fewer policy options.

In terms of the theory in section 2, a marginal factor of 100 percent on approximately 20

percent of all revenue raised spent in the states (in 1982) yields an α of just .2. Given the

comparative statics in section 2, this implies that Mexican states in the early 1980s faced very low

fiscal incentives to produce market fostering public goods. In recent years, the formula has changed,

but the situation has not improved markedly. We calculated that the proportion of revenue returned

to each state by the formula was .466. The revenue transferred to the states allocated by the formula

is now about one-half of all revenue transferred to the states. Taken together, these two observations

imply that, in present day Mexico, α, the marginal revenue return from expenditures fostering local

economic prosperity, is .466*.5 = .233.

We do not have the figures on total federal spending in the states prior to 1980, hence the

implied α is hard to assess for the earlier period. Yet the data in figure 6.2 and table 6.3 bear on this

36

issue. Federal taxes and spending in the states increased after 1980. With respect to state taxes: prior

to the final FPWD in 1972, states raised 45 percent of their own revenue; this proportion fell to 29

percent for the period 1973-80, and then dramatically fell again, to 3 percent for the period after the

1980 fiscal changes. These figures are average revenue raised, while α is the marginal revenue

collection. Nonetheless, the systematic changes in 1972 and in 1980, dramatically increasing federal

taxes and lowering state taxes, suggest that α fell significantly.

D. The Effects of Political Competition in Municipal Government

Our theory predicts that SNGs, which face electoral competition, have a stronger

preference for providing public goods over rents. In order to isolate the effects of political

competition from those of changes in α, we investigate the changes in municipalities after 1988,

when electoral competition became a reality but α remained constant. The findings accord with

our theory.

A voluminous case study literature has emerged in the last few years that studies the

changes in government operation in municipalities in Mexico.13 Many analysts say that these

changes amount to a “new municipal governance.” Ward (1998) summarizes the general trends

of change in municipal governance observed in these case studies. First, recruitment patterns for

both nominations to elected office and appointments have moved from a clientelistic system to

one based on the skills, experience, and local popularity of the prospective officers. Second,

municipalities have increased their tax collection efforts and have greatly diminished the

dependence on federal and state transferences, and to finance a high portion of their expenses

with municipal taxes. Third, the political agenda of municipalities has moved from being

13 See for example Rodríguez and Ward (1992, 1995, 1999); Merino (1994); Ziccardi A. (1995); Cabrero Mendoza (1995, 1996); Garza Villarreal (1998); and Guillén López (1995, 1996).

37

concerned only about the provision of basic services, to focusing on promoting rule of law and

markets, including: modernizing the police, promoting cleaner markets and slaughterhouses,

building highways, paving streets, and engaging in urban planning. Fourth, because political

competition implies that their political careers depend more on the popular perception of their

job than on the leadership of their parties, municipal presidents have become more autonomous

from their parties and from the corporatist organizations to which traditional PRI administrations

had very strong bonds.

These changes are more pronounced in the richer and more urbanized municipalities that

have more administrative capacity and possibilities to generate their own income. The changes

are stronger in Panista cities in the North and the rich Bajío region. The export-orientation of

many of these areas also increases the demand for the good governance necessary to promote

growing markets.

The fiscal experience of the municipalities electing the opposition PAN in the 1980s and

1990s accords with our theory. Diaz-Cayeros, Magaloni, and Weingast (2000) show that

municipalities that defect from the PRI by electing the opposition pay a substantial fiscal penalty

in terms of lower revenue transfers from the PRI controlled state and federal governments.

Our theory helps explain how the PAN survived despite the PRI’s fiscal punishment. As

a market-oriented municipality opted out of the PRI-fiscal system to take control of local public

goods provision, its revenue increased along with the local economy. In combination with

lowering levels of corruption, enhanced delivery of public goods and services greatly increased

local citizens’ willingness to pay taxes and new user fees. For localities that elected PAN

governments, local tax revenue grew quickly to replaces losses from the state and federal

governments. Rodriguez (1995,166) reports of Ciudad Juarez, for example, that ΑOver the

38

course of only a few years [after electing the PAN], the ratio of state to local revenues… changed

from around 70 percent state funding to over 70 percent local funding (ingresos propios).≅ In

particular, during the first year of the panista government, 1984, local revenue increased 300

percent.

Moreover, per our theory, the changes in behavior of municipal governments are not just

associated with the new opposition governments. These changes can also be seen where the PRI

municipal governments that face political competition. As with the opposition governments, PRI

governments facing competition are forced to improve services and the fiscal system.

E. The Effects of Revenue Sharing on Economic Growth

We argue that revenue sharing pacts motivate local politicians to reduce efficiency and

increase corruption, and therefore dampen economic growth. Mexico’s FPWD seems best

interpreted as helping the center create political power and rents rather than enhancing

efficiency. Our theory has several predictions about the impacts of changes in α over growth in

Mexico. In the first period (1917-1940) common pool problems slowed down growth. In the

Second Period (1947-1980) the ability of the center to police the common market brought about

major efficiency improvements. However, by coaxing states to join the FPWD, the center also

compromised growth in those states. Growth was dampened in the third period (1980-1994).

Since 1994, increases in both competition and in α have boosted growth. The figures discussed in

the introduction bear out this pattern: growth averaged 2.3 percent from 1980-93 per year,

increasing to 5.1 percent from 1996-99.

To study the effects of the fiscal changes on economic growth, we created a sample of

five states that joined the FPWD early and five that joined late. The fiscal system treated these

39

states differently, with significant effects on growth. The sample included: early joiners: Aguas

Calientes, Campeche, Guerrero, Michoacan, and Puebla; and late joiners: Chihuahua, Durango,

Mexico, Nuevo Leon, and Zacateca.

As noted above, late joining states were richer. For expenditures, expenditures in late

joining states averaged 647 million pesos per year (1970-72), while those joining early averaged

178 million, a ratio of about 3.6 (late to early). Similarly late joiners had larger state GDPs. For

the year 1970, early joiners average 7.5 billion pesos GDP while late joiners the figure is 18.1

billion pesos, about 2.4 times larger.

As we have already noted, two major events occurred in the early 1980s that affected

these patterns: the change in the fiscal system, resulting in greater centralization; and the

exogenous economic shocks caused the international debt crisis and by collapsing oil prices.

Given the data we have, we cannot separate the effects of these two changes.

The first major change is the relative growth of state expenditures among the early

joiners. We have calculated the ratio of state expenditures (late to early joiners) for the two

periods, 1970-80; and 1980-94. This calculation is reproduced in table 6.5.

Table 6.5: Change in the Ratio of State Expenditures (Late Joiners to Early Joiners) After 1980

Period 1970-80 1980-94

Ratio (late to early) 4.17 2.16

The data reveal a dramatic change in expenditures: the ratio of expenditures (late to early joiners)

falls by nearly a factor of one-half after 1980. Because nearly all state expenditures after 1980

40

were financed through the federal revenue sharing system, the new tax system clearly gave

disproportionately more revenue to the early joiners.

We next turn to the state GDP growth rates. By separating the states into the two sets of

early and late joiners, we are able to assess the differential impact of the two changes (changes in

the fiscal system; and exogenous economic shocks) across the two sets.

The data show that real state GDP growth rates for the late joiners is somewhat larger

than the early joiners in the early period, 7.5 percent versus 6.6 percent per year respectively

(table 6.4).

Table 6.4: Average Real State GDP Growth Rate, By Group and Period (percent per year)

1970-80 1980-93

Early Joiners 6.6 3.0

Late Joiners 7.5 1.7

The data also show that the growth rates of both groups fall dramatically after 1980, as is well-

known. The fall, however, is much larger for the late joiners: average GDP growth per year is 3.0

percent for the early joiners while only 1.7 percent for the late joiners.

The change in growth rates is consistent with the FPWD hypothesis: growth rates should

fall more in states joining later. Moreover, the smaller fall for the early joiners is consistent with

the evidence from the data on expenditures (and hence redistribution). Because the early joiners

get proportionately more transfers under the revenue sharing scheme, they are likely to growth at

a greater rate (assuming that some of the revenue is spent on public goods!). Of course, there are

many alternative hypotheses about why state GDP growth rates should fall. Perhaps the debt

41

crisis and the oil shock hit the richer states harder. Nonetheless, the fact that growth rates for the

late joiners fall more than for the early joiners is consistent with our hypothesis.

7. Conclusions

In this paper, we examine how political institutions affect good governance. Our context is fiscal

federalism. We show that different forms of decentralization have markedly different effects on

governance, as indicated by the incentives of SNGs to foster local economic prosperity.

The theory examines how SNGs make tradeoffs between providing public goods that

foster markets and non-productive expenditures, such as rent-seeking, transfers to interest

groups, and corruption. The analysis shows that revenue sharing greatly reduces a stateζs

incentive to produce public goods. When an SNG raises the lionζs share of its revenue, it has the

possibility of recouping the costs of providing market-fostering public goods through the

increased revenue generated. When an SNG derives most of their funds from a revenue sharing

system, increases in revenue from providing public goods go into the common pool, shared

among all states. Because an SNG bears all the costs of providing public goods but receives only

a portion of the return, it will provide a much lower level of public goods.

In comparison with engaging in corruption or rent seeking, providing public goods

generates two sources of value for an SNG. Because citizens value public goods, these goods

42

generate direct utility for citizens. Public goods have an indirect value as well: providing public

goods generates more tax revenue, thus relaxing the SNG’s budget constraint.

The model yields a comparative statics result. The greater the proportion of locally

generated revenue captured by the SNG, the more the SNG substitutes public goods provision

for corruption. We showed that a major consequence of this result is the fiscal law of 1/n: the

indirect effect of fostering greater SNG choice of public goods in a complete revenue sharing

system is 1/n that of a fiscal system in which SNGs capture all locally generated revenue. The

reason is that in revenue sharing systems that allocate revenue independent of contributions,

nearly all the increased revenue from providing additional public goods goes into the common

pool, implying that the SNG receives only on the order of 1/n of the increase. Revenue sharing

systems therefore greatly diminish an SNG’s incentive to provide market fostering public goods;

they also increase corruption and rent-seeking.

Consistent with this result is the following pattern among federal systems. During the

United States’s rise from a small economy on the periphery of the developed world in the late

eighteenth century to become the richest nation by early twentieth century, states depended

almost exclusively on their own sources of revenue. So too do provinces in modern China (see

Jin, Qian, and Weingast 2001, although this has changed somewhat in the late 1990s).

Iaryczower, Saiegh, and Tommasi (2000) suggest that this characteristic also held in Argentina

during its high growth phase in the latter part of the nineteenth century and early twentieth

centuries. In contrast, this condition fails for modern Argentina, India, Mexico, and Russia.

We applied our framework to Mexico, whose federal system has gone through four

phases since the Revolution in the early 20th century, each with different implications for

efficiency and growth. In the first federal phase, the absence of central control fostered common

43

pool problems as states over-grazed the commons. Too weak a center allowed competitive

taxation of businesses, state corruption, and internal trade barriers, resulting in significant

efficiency losses. In the second phase, beginning in 1940, the federal government rationalized

taxation, removed burdensome taxes, established the common market, and generally policed

common pool problems. States maintained several independent sources of revenue. High growth

ensued for the next thirty years, resulting in a period known as the ΑMexican Miracle.”

In the third phase of Mexican federalism (after 1980), the central government worked in

tandem with the ruling political party, the PRI, to centralize power, authority, and finances.

Gradually, the central government coaxed the states into giving up their policy and taxation

authority in exchange for revenue and attractive career alternatives for politicians. Growth stalled

substantially during the third phase.14 The theory implies that Mexico’s present revenue sharing

scheme is characterized by the fiscal law of 1/n.

Our analysis of the various revenue sharing formulas employed by the central

government bear out the theory. The initial formula returned almost exactly the same revenue to

the states as they put in. The purpose of this formula could not be either of the principal

normative rationales for revenue sharing -- redistribution or insurance against shocks -- there was

none. All the formula employed by the central government since 1980 exhibit aspects of the

fiscal law of 1/n.

This paper also analyzed the central governmentζs method of coaxing the states to join

the fiscal pact with the devil. These pacts forced states to toe the national party line, greatly

reducing their ability to pursue policies independent of the center. The central government

accomplished this by a series of carrots and sticks. On the cost side, the federal government

14 Of course, this third phase of Mexican federalism also coincided with the onset of the international debt crisis, which had deep effects on Mexico.

44

collected its tax revenue in each state regardless of whether they joined the revenue collection

system. On the benefit side, states obtained greater revenue if they joined the federal system. The

center’s incentives implied that retaining policy freedom forced states to have higher taxes and

lower revenue.

In closing, we speculate on a larger question. Our emphasis on the costs of revenue

sharing forces us to ask why the Mexican government created this system. The answer, we argue,

is that Mexico’s dominant ruling party, the PRI, created this system to help preserve its power.

First, the revenue sharing system afforded the PRI, via its control of the federal government, a

near monopoly on taxes. This allowed them to extract greater total revenue from the economy

than could the states acting independently. Second, the PRI used this revenue to solidify

constituent support. Nearly all empirical investigations of Mexico’s expenditure programs show

that they are used to help the PRI win elections rather than to address the policies to which they

are nominally associated (see Diaz-Cayeros 1997, Diaz-Cayeros, Magaloni, and Weingast 2000;

Magaloni 2000; Mortgenstern 19**; and Weldon and Molinar 1994). Third, the system helps the

PRI keep its officials in line with the party. By reducing the policy and fiscal authority of states,

the PRI made it difficult for SNG officials to simultaneously defect from the system and create

an independent source of political power. Finally, following Diaz-Cayeros, Magaloni, and

Weingast (2000) and Fiorina and Noll (1978), we argue that this system provides citizens with

the incentive to accept it. The reason is that, acting alone, citizens in a given locality cannot

affect the system and yet face direct punishment. Yet electing the opposition means that the

center punishes them, as the PRI withholds fiscal resources.

The “tragic logic” of modern Mexican political economy is that the PRI’s system at once

accomplished two goals: It skimmed a large portion of social resources used to ensure its own

45

survival; and it provided citizens with the incentive to go along (Diaz-Cayeros, Magaloni, and

Weingast 2000).

The model has implications beyond the study of Mexico. Our approach shows that good

governance requires the appropriately designed political institutions. For government officials to

implement policies that increase social surplus, they must have incentives to pursue these

policies instead of transferring resources to private groups or engaging in corruption (see also

Persson and Tabellini 2000, among others). We showed how fiscal institutions affect how

political officials make the tradeoff between the provision of public goods and corruption.

Greater fiscal reliance of SNGs on their own resources induces them to provide more public

goods and less corruption. The model shows that the degree of corruption and inefficiency is in

part endogenous to the fiscal system.

When central government policies are a major impediment to economic growth, the

appropriately structured decentralization can enhance government performance. First, as is well

known, appropriately structured decentralization creates competition among jurisdiction, forcing

them to attend to the inefficiencies, rent seeking and corruption associated with their

policymaking. Second, it diminishes the Αone-size fits all” problem observed by Hayek (1939).

Third, as we have shown in this paper, fiscal decentralization also provides SNGs with the

incentive to foster local economic prosperity.

In this work, we join a growing group of scholars who emphasize that good governance is

a function of political institutions. Our exploration of the effects of the fiscal system on SNG

decision-making demonstrates how fiscal institutions affect their decisions to foster markets or

engage in corruption. Put simply, greater revenue self-reliance allows SNGs the ability to capture

46

greater revenue from enhancing markets and thus biases their choices in favor of market

fostering public goods over corruption.

47

Appendix. Changes in the Allocation Criteria of the Mexican Revenue Sharing Formula

Since 1980. The following text box summarizes various changes since 1980 in the criteria utilized by the Mexican federal government to share revenue with the states.

In 1980 the Transfers fund was allocated among the states solely on the basis of the

proportion of federal taxes raised in each state. In 1981, each state received the same nominal amount as in the previous year, and the

increase in the Transfers fund (i.e. Transfers fund 1981- Transfers fund 1980) was distributed among the states taking into account the collection effort of each in the current year.

In 1983 the formula again assigned each state the same nominal amount as the

previous year, plus a proportion of the increase in the Transfers fund taking into account their collection efforts of the year with respect to the past two years.

In 1984, the central government attempted to stimulate the collection of the Value

Added Tax (IVA). Since the reform of 1980, states were responsible for collecting the IVA and for sending a portion to the federation. In this year, the formula allowed each state to keep 30% of the IVA it collected, plus an amount similar to that received by each the year before, modified a bit by general collection effort of the past year.

In 1990 the federation took IVA collection away from the states and made transfers

based on three criteria: population, tax collection effort, a small part inversely related to population. During the next four years, the federal government tinkered with the relative weight of the three criteria. Table 4.2 shows the percentage of the Transfers fund that was distributed by the three assignations criteria in the years 1991 to 1994.

Since 1994, 45.17% of the Transfers fund has been distributed according to state

population; 9.66% in inverse per capita relation; and 45.17% according to a formula that awards the states the same nominal amount as the previous year, plus a proportion of the increase in the Transfers fund taking into account their collection efforts of the year with respect to the past two years.

48

Since 1991, the revenue-sharing formula has been the same, consisting of three

components: population, revenue raised in the state, and a category representing the inverse of

the other two. However, the weights of the categories changed. The following table shows the

weights accorded to the categories since 1991.

Table A.1. Weights in the Distribution Criteria of the of the Transfer Fund by Year Year % distributed by

population % distributed by formula based on revenue collection

% distributed in inverse to the other two criteria

1991 18.05 72.29 9.66 1992 27.10 63.24 9.66 1993 36.15 54.19 9.66 1994-present 45.17 45.17 9.66 Source: Presidencia de la Republica

49

References Acemoglu, Daron, and James A. Robinson. 2000. “Why did the West Extend the Franchise?

Democracy, Inequality, and Growth in Historical Perspective, The Quarterly Journal of Economics 115 (4): 1167-1199

Alesina, Alberto, and Enrico Spolare. 1997. “On the Number and Size of Nations” Quarterly

Journal of Economics. 112(4): 1027-56 Alesina, Alberto, and Allan Drazen. 1991. “Why Are Stabilizations Delayed?” American

Economic Rev. 81: 1170-88. Bailey, John. 1995. “Fiscal Centralism and Pragmatic Accommodation in Nuevo Leon,” Ch 11

in Rodriguez and Ward (1995) Besley, Timothy, and Stephen Coate. 1998. “Sources of inefficiency in a representative

democracy: A dynamic analysis,” AER 88(1): 139-56. Bolton, Patrick, and Gerard Roland 1997. “The Breakup of Nations: A Political Economy

Analysis.” Quarterly Journal of Economics. 112(4): 1027-56 Brennan, Geoffrey, and James M. Buchanan. 1980. The Power to Tax. New York: Cambridge

University Press. Cabrero Mendoza, Enrique. 1996. Los Dilemas de la Modernización Municipal: estudios sobre la

gestión hacendaria en municipios urbanos de México. [México DF: CIDE] Cabrero Mendoza, Enrique. 1995. La Nueva Gestión Municipal en México: análisis de

experiencias inovadoras en gobiernos locales. [México DF: CIDE] Cao, Yuanzheng, Yingyi Qian, and Barry R. Weingast. 1999. “From Federalism, Chinese Style,

to Privatization, Chinese Style”Economics of Transition. vol 7 no. 1: 103-31. Cornelius, Wayne, Ann L. Craig, and Jonathan Fox. 1994. Transforming State-Society Relations

in Mexico: The National Solidarity Strategy. San Diego: Center for U.S.-Mexican Studies Cremer, Jacques, and Thomas Palfrey. Forthcoming. “Federal Mandates by Popular Demand,”

Journal of Political Economy. Cox, Gary, and Mathew D. McCubbins. 2001. “Institutions and Public Policy in Presidential

Systems,” in Stephan Haggard and Mathew D. McCubbins, eds., Presidents, Parliaments, and Policy. New York: Cambridge University Press.

50

Cornelius, Wayne A., Todd A. Eisenstadt, and Jane Hindley. 1999. Subnational Politics and Democratization in Mexico. La Jolla: Center for U.S. – Mexican Studies, University of California, San Diego.

Diaz-Cayeros, Alberto. “Political Responses to Regional Inequality: Taxation and Distribution in

Mexico,” Unpublished diss., Duke University, 1997. Diaz-Cayeros, Alberto. 1999. “Do Federal Institutions Matter? Rules and Political Practices in

Mexico,” W.P., Centro de Investigacion para el Dessarrollo, A.C. Mexico. 1999. Diaz-Cayeros, Alberto, and Beatriz Magaloni. 2000. “From Authoritarianism to Democracy: The

Unfinished Transition in Mexico.” W.P. Stanford University. Diaz-Cayeros, Alberto, Beatriz Magaloni, and Barry R. Weingast. 2000. “Democratization and

the Economy in Mexico: Equilibrium (PRI) Hegemony and its Demise.” W.P. Stanford University.

Dillinger, William, and Stephen B. Webb. 1999. “Fiscal Management in Federal

Democracies:Argentina and Brazil,” WP, World Bank. Dixit, Avinash, and John Londregan. 1995. “Redistributive Politics and Economic Efficiency,”

American Political Science Rev. Drazen, Allan. 2000. Political Economy in Macroeconomics. Princeton: Princeton University

Press. DeFiguieredo, Rui, and Barry R. Weingast. 2000. “Self-Enforcing Federalism: Solving the Two

Fundamental Dilemmas.” WP, Hoover Institution. Fiorina, Morris P., and Roger G. Noll. 1978. “Voters, Bureaucrats, and Legislators: A Rational

Choice Perspective on the Growth of Bureaucracy,” J. Pub. Economics (1978) 9: 239-54. Garman, Christopher, Stephan Haggard, and Eliza Willis. 2001. “Fiscal Decentralization: A

Political Theory with Latin American Cases.” World Politics 53(2): 205-36. Garza Villareal, Gustavo. 1998. La Gestión Municipal en el Area Metropolitana de Monterrey

1989-1994. [Mexico DF: Instituto de Investigaciones Sociales, UNAM] Geddes Barbara. 1999. “What do we know about Democratization after Twenty Years” Annual

Review of Political Science 2:115-144. Guillen López, Tonáhtiu. 1995. Municipios en Transición: actors sociales y nuevas políticas de

gobierno. [México DF: Ford Foundation] Guillen López, Tonáhtiu. 1996. Gobiernos Municipales en México: entre la moderniozación y la

tradición política. [Mexico: el Colegio de la Frontera Norte, Porúa]

51

Haughwout, Andrew, and Robert P. Inman. 2000a. “Fiscal Policies in Open Cities with Firms

and Households.” NBER WP 7823. Haughwout, Andrew, Robert P. Inman, Steve Craig, and Thomas Luce. 2000b. “Local Revenue

Hills: A General Equilibrium Specification with Evidence from Four U.S. Cities,” NBER WP 7603

Hayek, Frederich A. 1939. "The Economic Conditions of Interstate Federalism," New

Commonwealth Quarterly V (1939): 131-49, reprinted in his Individualism and Economic Order. Chicago: University of Chicago Press, 1948.

Hzou, Heng-fu. 1998. “Fiscal Decentralization and Economic Growth: A Cross-Country Study,”

Journal of Urban Economics, 43:244-257. Iaryczower, Matias, Sebastian Saiegh, and Mariano Tommasi. 2000. “Coming Together: The

Industrial Organization of Federalism.” WP, Universidad de San Andres, Argentina. Inman, Robert P. 1988. “Federal Assistance and Local Services in the United States: The

Evolution of a New Federalist Fiscal Order,” in Harvey S. Rosen, ed., Fiscal Federalism: Quantitative Studies. Chicago: University of Chicago Press.

Inman, Robert P, and Daniel L. Rubinfeld. 1997. “The Political Economy of Federalism,” in

Dennis C. Mueller, ed., Perspectives on Public Choice Theory. New York: Cambridge University Press.

Islas Torres, Alberto. 1997. The Political economy of Fiscal Decentralization: The case of

Mexico. MS Thesis, Political Science Department, Massachusetts Institute of Technology.

Jin, Hehui, Yingyi Qian, and Barry R. Weingast. 2000. “Regional Decentralization and Fiscal

Incentives: Federalism, Chinese Style,” Working Paper, Hoover Institution, Stanford University, March, 1999.

Jones, Mark P., Pablo Sanguinetti, and Mariano Tommasi. 2000. “Politics, Institutions, and

Fiscal Performance in a Federal System: An Analysis of the Argentine Provinces.” J. of Development Economics (forthcoming).

Knack, Stephen, and Philip Keefer. 1995. “Institutions and Economic Performance: Cross-

Country Tests using Alternative Institutional Measures,” Economics and Politics . (November) 7: 207-227.

Kraemer, Moritz. 1997. “Intergovernmental Transfers and Political Representation: Empirical

Evidence from Argentina, Brazil, and Mexico.” Inter-American Development Bank, Working Paper 345. (January)

52

Levy, Brian, and Pablo T. Spiller. "Institutional Foundations of Regulatory Commitment: A Comparative Analysis of Telecommunications Regulation," Journal of Law, Economics, and Organization (1994) 10: 201-46.

Londregan, John, and Pool Keith. 1990. “Poverty, the Coup Trap and the seizure of Executive

Power” World Politics. 42:151-183. Lujambio Alonso. 2000. El Poder Compartido: un ensayo sobre la democratización mexicana.

[Mexico DF: OCEANO] Magaloni, Beatriz. 2000. “Institutions, Political Opportunism, and Macroeconomic Cycles:

Mexico, 1970-1998,” WP, Stanford University Merino, Mauricio. 1994. En Busca de la Democracia Municipal: la participación ciudadana en el

gobierno local mexicano [México DF: COLMEX]

McKinnon, Ronald I. 1997. “Market-Preserving Fiscal Federalism in the American Monetary Union,” in Mario I. Blejer and Teresa Ter-Minassian, eds., Macroeconomic Dimensions of Public Finance, New York: Routledge, pp.73-93.

McKinnon, Ronald, and Thomas Nechyba. 1997. “Competition in Federal Systems: Political and

Financial Constraints,” in John Ferejohn and Barry R. Weingast, eds., The New Federalism: Can the States be Trusted? Stanford: Hoover Institution Press.

Molinar Horcasitas, Juan, and Jeffrey A. Weldon. 1994. “Electoral Determinants and

Consequences of National Solidarity,” in Cornelius, et. Al. (1994) Montinola, Gabriella, Yingyi Qian, and Barry R. Weingast. 1995. “Federalism, Chinese Style.”

World Politics. 48(1): 50-81. North, Douglass C. 1990. Institutions, Institutional Change and Economic Performance. New

York: Cambridge University Press. Oi, Jean. 1999. Rural China Takes Off: Institutional Foundations of Economic Reform. Berkeley:

University of California Press. Parikh, Sunita, and Barry R. Weingast. 1997. “A Comparative Theory of Federalism: India.”

Virginia Law Rev. 83(7): 1593-1615. Persson, Torsten, and Guido Tabelini. 1993. “A Federal Fiscal Constitutions: Risk Sharing and

Redistribution," J. of Political Economy Persson, Torsten, and Guido Tabellini. 2000. Political Economics: Explaining Economic Policy.

Cambridge: M.I.T. Press.

53

Poterba, James M., and Jurgen von Hagen. 1999. Fiscal Institutions and Fiscal Performance. Chicago: University of Chicago Press.

Presidencia de la Republica 1998. Prud'homme, Remy. 1995. “On the Dangers of Decentralization.” World Bank Research

Observer. 10,2: 201-20. Przeworski, Adam, and Limongy Fernando. “Modernization: Theories and Facts”. World

Politics. 49:155-83. Riker, William H. 1964. Federalism. Boston: Little, Brown. Robinson, James C. 1999. "Why Did the West Extend the Franchise? Growth, Inequality and

Democracy in Historical Perspective," Quarterly Journal of Economics 115: 1150-1171. Rodden, Jonathan. 2000. “The Dilemma of Fiscal Federalism: Hard and Soft Budget Constraints round

the World.” W.P. M.I.T. Rodriguez , Victoria E. 1995. “Municipal Autonomy and the Politics of Inter governmental

Finance: Is it Different for the Opposition?” Ch 10 in Rodriguez and Ward (1995) Rodriguez, Victoria E. 1997. Decentralization in Mexico: From Reforma Municipal to

Solidaridad to Nuevo Federalismo. Boulder: Westview Press. Rodriguez, Victoria, and Ward Peter. 1992. Policymaking, Politics and Urban Governance in

Chihuahua: the experience of recent Panista government. [Austin TX: LBJ School of Public Affairs, UT]

Rodriguez, Victoria E., and Peter M. Ward, eds., Opposition Government in Mexico.

Albuquerque: University of New Mexico Press, 1995. Rodriguez, Victoria E., and Peter M. Ward. 1999. The New Federalism and State Government in

Mexico: Bringing the States Back In. University of Texas, Austin. Rodrik, Dani. 1999. “Institutions for High-Quality Growth: What Are They and How To Acquire

Them?” Working Paper, Harvard University. Rubinfeld, Daniel. 1987. "Economics of the Local Public Sector," Handbook of Public

Economics vol. II, ed. by A. J. Auerbach and M. Feldstein. New York: Elsevier. Roland, Gerard. 2001. Transition and Economics. Politics, Markets, and Firms. Cambridge:

M.I.T. Press. Saez, Lawrence. 1998. “A Comparison of India and China’s Foreign Investment Strategy toward

Energy Infrastructure,” J. of Developing Areas 32 (Win): 199-220.

54

Sanguinetti, Pablo J. 1994. “Intergovernmental Transfers and Public Sector Expenditures: A Game-Theoretic Approach.” Estudios de Economia. 21(2): 179-212.

Shah, Anwar. 1997. “Balance, Accountability, and Responsiveness: Lessons from

Decentralization.” WP #, World Bank, Washington, D.C. Shirk, Susan. 1993. The Political Logic of Economic Reform in China. Berkeley: University of

California Press. Stein, Ernesto. 1998. “Fiscal Decentralization and Government Size in Latin America,” pp. 95-

119. In Kiichiro Fukasako and Ricardo Hausmann, eds., Democracy, Decentralization and Deficits in Latin America. Washington, D.C.: IADB-OECD.

Stiglitz , Joseph. 1998. “Distinguished Lecture on Economics in Government: The Private Uses

of Public Interests: Incentives and Institutions.” J. of Economic Perspectives 12(2): 3-22. Sturzenegger, Federico, and Mariano Tommasi. 1998. Political Economy of Reform. Cambridge:

M.I.T. Press. Tiebout, Charles. 1956.”A Pure Theory of Local Expenditures,” Journal of Political Economy,

October, 64, pp.416-24. Treisman, Daniel. 1999. “Decentralization and Inflation: Commitment, Collective Action, or

Continuity?” Working Paper, UCLA. Ward, Peter M. 1995. “Policymaking and Policy Implementation among Non-PRI Governments:

The PAN in Ciudad Juarez and in Chihuahua,” Ch 9 in Rodriguez and Ward (1995) Weingast, Barry R. 1995. "The Economic Role of Political Institutions: Market-Preserving

Federalism and Economic Development." Journal of Law, Economics, and Organization (Spring) 11: 1-31.

Weingast, Barry R. 2000. “The Comparative Theory of Federalism.” WP, Hoover Institution,

Stanford University. Weingast, Barry R., Kenneth A. Shepsle, and Christopher Johnsen. 1981. “Political Economy of

Benefits and Costs.” Journal of Political Economy. Weldon, Jeffrey. 1997. “The Political Sources of Presidencialismo in Mexico,” in Scott

Mainwaring and Matthew Soberg Shugart, Presidentialism and Democracy in Latin America. New York: Cambridge University Press.

Wildasin, David E. 1997. "Externalities and Bailouts: Hard and Soft Budget Constrains in

Intergovernmental Fiscal Relations.” mimeo, Vanderbilt University

55

Willis, Eliza, Christopher Garman, and Stephan Haggard. 1998. “The Politics of Decentralization in Latin America,” WP, UCSD.

Zaid Gabriel. 2000. La Economia Presidencial. Mexico DF: Oceano. Zhuravskaya Ekaterina V. 2000. “Incentives to provide local public goods: fiscal federalism,

Russian style,” Journal Of Public Economics (76)3: 337-368 Zicardi, Alicia. 1995. La Tarea de Gobernar: gobiernos locales y demandas ciudadanas. [México

DF: Instituto de Investigaciones Sociales, UNAM]